agreement conflict? (from Michael Daniel) (fwd)

Edith A Moravcsik edith at CSD.UWM.EDU
Mon Jan 25 15:11:11 UTC 1999

			 Edith A. Moravcsik
			 Department of Foreign Languages and Linguistics
			 University of Wisconsin-Milwaukee
		         Milwaukee, WI 53201-0413

			 E-mail: edith at
		         Telephone: (414) 229-6794 /office/
				    (414) 332-0141 /home/
		         Fax: (414) 229-2741


---------- Forwarded message ----------
Date: Fri, 22 Jan 1999 13:16:19 +0300
From: Michael Daniel <aroginskii at>
To: Edith A Moravcsik <edith at CSD.UWM.EDU>
Subject: Re: agreement conflict? (from Michael Daniel)

Dear Edith!

You were right in pointing that the morphological explanation of
conflicting agreement marking is excessively language specific, because
it does not explain the fact that "Nick's Pete's picture" is not allowed
in English. I think different explanations should be suggested for
English and Hungarian. In addition, I think that we should put the
question in a slightly different way, not:

because "Pete's father's house" seems to be OK in English.

I think, the situation in English may be given a functional explanation
(kind of Martin Haspelmath was looking for). The fact that in English
there can not be two preposed possessors of the same NP seems to be
connected to the fact that otherwise several readings would become
equally possible which correspond to different syntactic parsing of the
sequence, as follows:

A.    [my [family's branch]]
B.    [[my family]'s branch]
C.    [my family's branch]
(By not putting the second pair of brackets in the last parsing I meant
that both possessors are immediately connected to the head noun - no
group dependances; the last parsing is also dubious because the
semantics of nominal relations should be different, which is not always

In fact only the second parsing is allowed. It seems quite logical -
instead of having ambiguous sequence, English introduces the rule by
which the possessor NP may not be related to a complex NP which includes
several nouns (which excludes reading A) neither go overhead another
noun (which excludes reading C), but seeks rightward for the first
available noun and is then connected to it. In my opinion, it is only a
way to avoid possible ambiguity. For the two other readings English uses
other constructions - like NP's NP of NP for the reading A ("John's
share of the estate") or "NP and NP's NP" for simultaneous possession,
which is one of the readings of parsing C ("king and queen's palace").

If I understand correctly, the same rule holds true for some actualizers
(like unambiguous "The rear left horse's hoof"). One should admit,
however, that this rule is overgeneralized in English, because sentences
like "the left Kate's eye" or "my Pete's picture", that would cause no
ambiguity, yet do not allow nominal relations go over another noun and
are simply ungrammatical (or may they be understood as choosing between
several Kates and Petes?)

It seems to me that adnominal constructions like "John's estate share"
are allowed, because there are no overt markers of nominal relationship
between "estate" and "share". One may argue now that in English unmarked
relationship has a priority over marked relationship; I would add, that
unmarked nominal relationship seems to be more "tight" also
crosslinguistically, thus putting constructions of the type "estate
share" in between syntactic construction and morphological compound, and
there seems to be several restrictions on the referential status of
"unmarked possessor". In addition, simple nouns rather than NPs are
allowed in such constructions. If the other way of parsing would be
tried we would come to:
[[John's estate] share]
which would contradict these restrictions.

It is not evident, however, that such avoidance of ambiguity should be
universal. Language are thinkable of which would allow several readings
for 'NP-Gen NP-Gen NP' constructions, leaving the sequence ambiguous.
Thus I don't know if there is consistent priority of "of" relation over
's-relation (or vice versa) in English NP's NP of NP - constructions.

The situation in Russian seems to throw some light on it. The genitival
possessor is normally postposed to the possessed item; there is also
another way of expressing possession for animates - by possessive form
of the noun, which is a kind of an adjective, but unlike ordinary
adjectives may preserve the referentiality of the noun stem.
a.    stikhotvorenije Pushkin-a
        poem Pushkin-Gen
b.    Pushkin-sk-oje stikhotvorenije
        Pushkin-Poss-NomNtr poem

I don't think there is a consistent avoidance of two genitives
constructions with genitives; thus, the following sentence does not
really seem ungrammatical, though clumsy:

c.    [portret Pushkin-a [kist-i [neizvestn-ogo khudozhnik-a]].
        picture Pushkin-Gen brush-Gen unknown-GenMasc painter-Gen
    'Pushkin's portrait made by an unknown master'

However, constructions like:

d.  [chast' [nasledstv-a John'a]]
      part     estate-Gen John-Gen
    'part of John's estate'
are only understood as recursive genitives, not as 'John's part of the
estate'. To express the latter meaning, Russian would also use two
different ways of expressing nominal relations, exactly like English:
    Mish-in-a chast' nasledstv-a
    Michael-Poss-Gen part estate-Gen

It is not quite evident from your communication how exactly things are
in Hungarian; it would be great if you could provide some glossed
By saying that "family"/"estate" may become an adnominal modifier, do
you mean that it looks like examples from English or Turkish, where noun
stem is closely connected to the possessed item, forming a kind of

The fact that actualizers are allowed with possessed items in Hungarian
seems due to the fact that such actualizer is immediately preposed to
the actulized noun (am I right?), while in English it should go to the
phrase initial position.
H:     NP Art NP-Poss 'this house of (my) father
E:    Art NP's NP
Thus in English again several parsings become available, which is
avoided by introducing the rule analogous to that mentioned above
(looking rightward for the first noun stem available). In Hungarian
there are no grounds for syntactical (and thus semantical) ambiguity,
because the article position itself is disambiguating.

Michael Daniel

More information about the Lingtyp mailing list